FSC: Two-Scale AI-Driven Fair Sensitivity Control for 802.11ax Networks

With the excessive demand on the mobile and wireless usage, IEEE 802.11ax is released to enable concurrent transmissions in the dense WLANs, enhanced with the Spatial Reuse (SR) techniques. However, IEEE 802.11ax officially has not adaptive physical carrier sensing mechanism to be used with the SR techniques. The lack of adaptive carrier sensing leads to hidden/exposed nodes problems in dense networks. Also, it induces unfair access between the stations. Therefore, thorough the study, we try to solve how to design an adaptive carrier sensing mechanism to balance hidden/exposed problems and increase fairness for all stations in the dense WLANs. Consequently, we propose a two-scale Fair Sensitivity Control (FSC), which operates with both local-scale and global-scale to adjust the Carrier Sensitivity Threshold (CST) for stations. In contrast to other studies, we use local scale control to adaptively adjust CST to decrease interference and global-scale control using Artificial Intelligence (AI) to decrease the fairness issue caused by stations' placements. Thanks to the learning capabilities of the AI, specifically Multilayer Perceptron (MLP), that we have implemented in this work, the WLANs learn the particular carrier sensing parameter, called Margin for stations. We evaluate the FSC mechanism under the five metrics: i) aggregated throughput, (ii) number of collisions, (iii) number of hidden terminals, (iv) number of exposed terminals, and (v) fairness. Simulation results show a clear performance improvement over the five metrics compared with the current state of the art.